Harm without limits: AI child sexual abuse material through the eyes of our Analysts
This report provides an updated overview of the growing use of generative AI to create child sexual abuse material (“AI CSAM”), highlighting increases in scale, realism and accessibility, alongside legal and enforcement challenges.
Details
The report confirms a sharp escalation in AI-generated abuse imagery. In 2025, the Internet Watch Foundation (IWF) assessed 8,029 AI-generated images and videos depicting child sexual abuse, including 3,443 videos, a 26,385% increase on the previous year.
AI-generated material is now highly photorealistic and often difficult to distinguish from real abuse imagery, with 65% of videos classified as Category A – the highest level of severity. This reflects rapid advances in generative tools, such as fine-tuning techniques that enable users to produce bespoke content, including images of identifiable children, with minimal expertise.
Accessibility is also increasing. Single applications can generate images, video and audio content, lowering barriers to entry and enabling use by individuals with little technical knowledge. AI CSAM has been identified across both dark web and mainstream platforms, including chatbots, and is linked to grooming and sextortion.These developments are reshaping the nature, scale and accessibility of offending.
Against that backdrop, the report emphasises that AI CSAM is not “victimless”. Existing abuse material may be used to train AI models, resulting in ongoing revictimisation, while the availability of increasingly realistic content may reinforce sexual interest in children and contribute to escalation in offending.
These harms are compounded by growing operational challenges. Law enforcement must distinguish real and synthetic material, often at scale, and is beginning to encounter “deepfake defence” arguments, where offenders claim that genuine evidence of contact abuse is artificially generated.
In the UK, legislation already criminalises the creation, possession and distribution of indecent images of children, including AI-generated and non-photorealistic content. Further offences are due to come into force on 29 June 2026 with the Crime and Policing Act 2026, including criminalising the creation of ‘deepfakes’ of adults and the possession of AI models which can create CSAM. You can see further details of this in YJLC’s legal guide on Image Offences and Online Sexual Offending.
Commentary
For youth justice practitioners, the report highlights several issues.
First, children may increasingly be involved in offences linked to AI tools, particularly in peer-on-peer contexts (eg. “nudify” apps). Low technical barriers mean children may engage without fully understanding that the software or the content is unlawful. As a result, diversion is likely to be an appropriate response.
Secondly, practitioners should be aware that AI-generated content is used in grooming and extortion, meaning children may present as suspects while also having been coerced. Practitioners should consider safeguarding obligations and the appropriateness of prosecution where there are concerns about the child suspect or defendant having been groomed.
Thirdly, evidential challenges are likely to intensify. As synthetic material becomes harder to distinguish, questions of authenticity will be central, requiring early engagement with expert evidence.
Finally, broader child rights concerns arise. Girls are disproportionately targeted (97% of identified imagery), and existing abuse material is reused to generate further content, extending harm to known victims. Practitioners should reflect these issues in mitigation and welfare-focused submissions.